System, methods, and other embodiments described herein relate to identifying changes between models of a locality. In one embodiment, a method includes, in response to determining that a location model is available for a present environment of a vehicle, generating a current model of the present environment using at least one sensor of the vehicle. The method also includes isolating dynamic objects in the current model as a function of the location model. The method includes providing the dynamic objects to be identified and labeled.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A comparative labeling system for recognizing changes in locality models of a mapped environment, comprising: one or more processors; a memory communicably coupled to the one or more processors and storing: a mapping module including instructions that when executed by the one or more processors cause the one or more processors to, in response to determining that a location model is available for a present environment of a vehicle, generate a current model of the present environment using at least one sensor of the vehicle; and a comparison module including instructions that when executed by the one or more processors cause the one or more processors to: isolate dynamic objects in the current model as a function of the location model, wherein the dynamic objects include semi-permanent objects of the present environment that change between the current model and the location model, and provide the dynamic objects from the current model to be identified and labeled, wherein the location model is a three-dimensional representation of the present environment that was previously acquired and labeled to identify the semi-permanent objects including at least static aspects of the present environment.
This system detects and labels changes in a mapped environment, particularly for autonomous vehicles. The technology addresses the challenge of identifying dynamic objects in a vehicle's surroundings by comparing real-time sensor data with a pre-existing 3D map of the environment. The system uses one or more processors and memory to execute two key functions. First, a mapping module generates a current model of the environment using vehicle sensors when a location model (previously acquired 3D map) is available. Second, a comparison module isolates dynamic objects in the current model by comparing it to the location model. These dynamic objects include semi-permanent features that change between the current and pre-existing models, such as parked cars, temporary barriers, or construction zones. The system then provides these dynamic objects for identification and labeling, allowing the vehicle to recognize and adapt to environmental changes. The location model serves as a reference, containing static aspects of the environment (e.g., roads, buildings) and previously labeled semi-permanent objects. This approach improves situational awareness for autonomous navigation by distinguishing between permanent and temporary features.
2. The comparative labeling system of claim 1 , wherein the comparison module further includes instructions to isolate the dynamic objects by projecting the current model onto the location model to align the current model with the location model and to identify changes between the location model and the current model that correspond with the dynamic objects, and wherein the comparison module further includes instructions to isolate the dynamic objects including instructions to label mobile objects that are part of the dynamic objects as provisional when the mobile objects are presently not moving but are determined to likely move and to label the dynamic objects that are semi-permanent objects according to an object class.
This invention relates to a comparative labeling system for identifying and classifying dynamic objects in a monitored environment. The system addresses the challenge of accurately distinguishing between static and dynamic elements in an environment where objects may change position or state over time. The system includes a comparison module that processes a current model of the environment and a location model representing a reference state. The module aligns the current model with the location model by projecting the current model onto the location model, enabling the identification of changes between the two models. These changes correspond to dynamic objects, which may include mobile objects or semi-permanent objects. The system further classifies these dynamic objects: mobile objects that are not currently moving but are likely to move are labeled as provisional, while semi-permanent objects are labeled according to their object class. This classification helps in tracking and managing dynamic elements in the environment, improving situational awareness and decision-making. The system enhances accuracy in object detection and classification by leveraging model alignment and dynamic object isolation techniques.
3. The comparative labeling system of claim 2 , wherein the comparison module further includes instructions to project the current model onto the location model by aligning a perspective of the current model with the location model to overlay corresponding points between the current model and the location model, and wherein the comparison module further includes instructions to isolate the dynamic objects by comparing the current model with the location model to produce observation data points that embody the dynamic objects of the present environment while removing static objects that form a background of the present environment that is unchanged.
This invention relates to a comparative labeling system for identifying dynamic objects in an environment by comparing a current model of the environment with a pre-existing location model. The system addresses the challenge of distinguishing moving or changing objects from static background elements in real-time applications such as surveillance, autonomous navigation, or environmental monitoring. The system includes a comparison module that projects the current model onto the location model by aligning their perspectives to overlay corresponding points between them. This alignment ensures accurate spatial correlation between the two models. The comparison module then isolates dynamic objects by comparing the current model with the location model, generating observation data points that represent only the dynamic objects while filtering out static background elements that remain unchanged. This process effectively separates transient or moving objects from the static environment, enabling precise tracking and analysis of dynamic elements. The system leverages the location model as a reference to identify changes in the environment, improving accuracy in detecting and labeling dynamic objects. This approach enhances situational awareness in applications requiring real-time object detection and classification.
4. The comparative labeling system of claim 1 , wherein the location model is a three-dimensional representation of the present environment that was previously acquired and labeled to identify static objects in the present environment.
This invention relates to a comparative labeling system for identifying and labeling objects in a three-dimensional environment. The system addresses the challenge of accurately recognizing and categorizing static objects within a given space, particularly in applications such as robotics, augmented reality, or autonomous navigation where environmental awareness is critical. The system utilizes a three-dimensional representation of the current environment, which has been previously acquired and annotated to mark static objects. This pre-labeled model serves as a reference for comparing and updating object labels in real-time. The system dynamically adjusts labels based on sensor data, ensuring consistency between the pre-labeled model and the current environment. This approach improves object recognition accuracy by leveraging prior knowledge of the environment, reducing the need for real-time labeling and enhancing system efficiency. The three-dimensional representation includes detailed spatial and contextual information about static objects, allowing the system to distinguish between permanent fixtures and transient elements. By comparing the current sensor data with the pre-labeled model, the system can detect discrepancies, such as missing or relocated objects, and update labels accordingly. This ensures that the system maintains an accurate and up-to-date understanding of the environment, supporting applications requiring precise object identification and tracking. The system is particularly useful in dynamic environments where static objects may change over time, providing a robust framework for maintaining environmental awareness.
5. The comparative labeling system of claim 1 , wherein the comparison module further includes instructions to provide the dynamic objects by transmitting data from a light detection and ranging (LIDAR) sensor and a camera sensor about the dynamic objects to a remote server for labeling, and wherein the data from the camera sensor is validation data.
This invention relates to a comparative labeling system for dynamic objects in autonomous vehicle environments. The system addresses the challenge of accurately identifying and labeling moving objects such as vehicles, pedestrians, or obstacles in real-time using sensor data. The system includes a comparison module that processes data from multiple sensors, including a LIDAR sensor and a camera sensor, to generate and validate labels for dynamic objects. The LIDAR sensor provides spatial and distance measurements of the objects, while the camera sensor captures visual data used to validate the labels generated from the LIDAR data. The comparison module transmits this sensor data to a remote server, which performs the labeling process. The remote server analyzes the LIDAR data to identify and classify dynamic objects, then uses the camera data to verify the accuracy of these labels. This dual-sensor approach improves labeling reliability by cross-referencing spatial and visual information, reducing errors in object detection and classification. The system is particularly useful in autonomous driving, where precise object recognition is critical for navigation and safety.
6. The comparative labeling system of claim 1 , wherein the mapping module further includes instructions to generate the current model by scanning the present environment of the vehicle using at least a light detection and ranging (LIDAR) sensor to acquire the current model as a three-dimensional point cloud, and wherein comparison module includes instructions to provide the dynamic objects including instructions to electronically communicate at least LiDAR data of the dynamic objects to a remote system for labeling to improve updating of the location model using the dynamic objects.
This invention relates to a comparative labeling system for vehicles, specifically addressing the challenge of accurately updating location models in dynamic environments. The system includes a mapping module that generates a current model of the vehicle's environment by scanning the surroundings using a light detection and ranging (LiDAR) sensor, producing a three-dimensional point cloud representation. This module enables real-time environmental mapping for navigation and object detection. The system also features a comparison module that identifies dynamic objects within the environment. The comparison module further includes instructions to transmit LiDAR data of these dynamic objects to a remote system for labeling. This remote labeling process enhances the accuracy of the location model by incorporating real-time updates from dynamic objects, such as moving vehicles or pedestrians. The system leverages cloud-based processing to improve the precision of environmental mapping, ensuring that the vehicle's navigation and object detection systems remain up-to-date with the latest environmental data. This approach enhances situational awareness and safety in autonomous or semi-autonomous driving scenarios.
7. The comparative labeling system of claim 1 , wherein the vehicle is an autonomous vehicle.
The invention relates to a comparative labeling system designed for autonomous vehicles to enhance situational awareness and decision-making. The system compares real-time sensor data from the autonomous vehicle with pre-existing data to generate comparative labels that describe the vehicle's environment. These labels help the autonomous vehicle identify and classify objects, obstacles, and other relevant features in its surroundings. By analyzing differences between the real-time data and the pre-existing data, the system can detect changes in the environment, such as new obstacles or altered road conditions, and update the vehicle's perception of its surroundings. This improves the vehicle's ability to navigate safely and make informed decisions. The system may use various types of sensor data, including but not limited to, LiDAR, radar, and camera data, to generate accurate and reliable comparative labels. The invention aims to address the challenges of real-time environmental perception in autonomous vehicles, ensuring safer and more efficient operation.
8. The comparative labeling system of claim 1 , wherein the mapping module further includes instructions to: in response to determining the location model is not available for the present environment, generate the location model by scanning the present environment using the at least one sensor to obtain three-dimensional data points representing present objects in the present environment that are collected to form the location model, and label the present objects to identify types of the present objects of the present environment.
This invention relates to a comparative labeling system for identifying and categorizing objects in an environment using sensor data. The system addresses the challenge of accurately labeling objects in dynamic or unfamiliar environments where pre-existing location models are unavailable. The system includes a mapping module that generates a location model by scanning the environment with sensors to collect three-dimensional data points representing objects. These objects are then labeled to identify their types, enabling the system to recognize and categorize them within the environment. The system also compares the generated labels with reference labels from a database to verify accuracy and improve future labeling tasks. The mapping module dynamically adapts to new environments by generating location models on-the-fly when existing models are not available, ensuring consistent object recognition and labeling. This approach enhances the system's reliability in diverse settings, such as robotics, augmented reality, or autonomous navigation, where real-time object identification is critical. The system's ability to autonomously create and refine location models improves its adaptability and accuracy in dynamic environments.
9. A non-transitory computer-readable medium storing instructions that when executed by one or more processors cause the one or more processors to: in response to determining that a location model is available for a present environment of a vehicle, generate a current model of the present environment using at least one sensor of the vehicle; isolate dynamic objects in the current model as a function of the location model, wherein the dynamic objects include semi-permanent objects of the present environment that change between the current model and the location model; and provide the dynamic objects to be identified and labeled, wherein the location model is a three-dimensional representation of the present environment that was previously acquired and labeled to identify the semi-permanent objects including at least static aspects of the present environment.
This invention relates to autonomous vehicle navigation and environmental perception. The problem addressed is accurately identifying and labeling dynamic objects in a vehicle's environment while distinguishing them from static or semi-permanent features. The solution involves using a pre-existing three-dimensional location model of the environment, which includes labeled static and semi-permanent objects, to compare against real-time sensor data from the vehicle. The system generates a current model of the environment using vehicle sensors, then isolates dynamic objects by comparing the current model with the location model. Dynamic objects are defined as semi-permanent features that have changed between the current and location models. These dynamic objects are then provided for identification and labeling, allowing the vehicle to update its understanding of the environment. The location model serves as a reference to distinguish between static infrastructure and temporary or moving obstacles, improving navigation and safety. The approach leverages prior knowledge of the environment to enhance real-time perception accuracy.
10. The non-transitory computer-readable medium of claim 9 , wherein the instructions to isolate the dynamic objects include instructions to project the current model onto the location model to align the current model with the location model and to identify changes between the location model and the current model that correspond with the dynamic objects, and wherein the instructions to isolate the dynamic objects include instructions to label mobile objects that are part of the dynamic objects as provisional when the mobile objects are presently not moving but are determined to likely move and to label the dynamic objects that are semi-permanent objects according to an object class.
This invention relates to computer vision and object tracking, specifically for identifying and classifying dynamic objects in an environment. The problem addressed is distinguishing between static, semi-permanent, and mobile objects in a scene, particularly when some objects may appear stationary but are likely to move. The system uses a location model representing a known environment and a current model derived from real-time sensor data. The current model is aligned with the location model to detect differences, which correspond to dynamic objects. Mobile objects that are temporarily stationary but expected to move are labeled as provisional, while semi-permanent objects (e.g., furniture, equipment) are classified based on their object class. This approach improves object tracking by dynamically adjusting labels based on movement likelihood and permanence, enhancing accuracy in applications like autonomous navigation or surveillance. The method involves projecting the current model onto the location model to identify changes, then classifying the detected objects accordingly. The system differentiates between truly static elements and objects that may move later, reducing false negatives in tracking. The solution is implemented via software instructions stored on a non-transitory computer-readable medium, ensuring real-time processing and adaptability to varying environments.
11. The non-transitory computer-readable medium of claim 10 , wherein the instructions to project the current model onto the location model include instructions to align a perspective of the current model with the location model to overlay corresponding points between the location model and the current model, and wherein the instructions to isolate the dynamic objects include instructions to compare the current model with the location model to produce observation data points that embody the dynamic objects of the present environment while removing static objects that form a background of the present environment that is unchanged.
This invention relates to computer vision and environmental modeling, specifically for distinguishing dynamic objects from static backgrounds in real-time environments. The technology addresses the challenge of accurately identifying and tracking moving objects in a scene while filtering out static elements that form the unchanged background. The system uses a location model representing the static environment and a current model representing the observed scene. The current model is projected onto the location model by aligning their perspectives to overlay corresponding points, ensuring spatial consistency. By comparing the current model with the location model, the system generates observation data points that isolate dynamic objects while removing static objects, effectively separating moving elements from the background. This approach enhances object detection accuracy in applications such as surveillance, autonomous navigation, and augmented reality by dynamically filtering out irrelevant static structures. The method leverages model alignment and differential analysis to achieve real-time environmental awareness, improving the reliability of object tracking in varying conditions.
12. The non-transitory computer-readable medium of claim 9 , wherein the location model is a three-dimensional representation of the present environment that was previously acquired and labeled to identify static objects in the present environment, wherein the instructions to provide the dynamic objects include instructions to transmit data from a light detection and ranging (LIDAR) sensor and a camera sensor about the dynamic objects to a remote server for labeling, and wherein the data from the camera sensor is validation data.
This invention relates to systems for mapping and labeling dynamic objects in an environment using sensor data. The problem addressed is the need to accurately identify and track moving objects in real-time while maintaining a reliable representation of static elements in the environment. The solution involves a non-transitory computer-readable medium storing instructions for generating a three-dimensional representation of the environment, which has been previously acquired and labeled to identify static objects. The system further includes instructions for detecting and transmitting data about dynamic objects from both a LIDAR sensor and a camera sensor to a remote server for labeling. The camera sensor data serves as validation data to ensure accuracy in the labeling process. The LIDAR sensor provides precise spatial information about the dynamic objects, while the camera sensor captures visual data to verify and refine the labels assigned by the remote server. This approach improves the reliability of object detection and tracking in dynamic environments by combining multiple sensor inputs and leveraging remote processing for labeling. The system ensures that static objects remain accurately mapped while dynamically updating the environment with labeled moving objects.
13. The non-transitory computer-readable medium of claim 9 , wherein the instructions to generate the current model include instructions to scan the present environment of the vehicle using at least a light detection and ranging (LIDAR) sensor to acquire the current model as a three-dimensional point cloud.
This invention relates to autonomous vehicle navigation systems that use environmental scanning to generate real-time models of the vehicle's surroundings. The problem addressed is the need for accurate, up-to-date 3D representations of the environment to support safe and efficient autonomous driving. The system employs a LIDAR sensor to scan the vehicle's current environment, capturing detailed spatial data as a three-dimensional point cloud. This point cloud serves as the current model, providing precise geometric and positional information about objects, terrain, and obstacles in the vicinity. The LIDAR-based scanning ensures high-resolution, real-time data acquisition, which is critical for dynamic decision-making in autonomous driving scenarios. The system integrates this model with other navigation and control processes to enhance situational awareness and path planning. The use of LIDAR ensures robustness against varying lighting conditions and provides fine-grained details necessary for obstacle detection, collision avoidance, and precise localization. This approach improves the reliability and safety of autonomous vehicle operations by continuously updating the environmental model to reflect real-world conditions.
14. A method of identifying differences between locality models, comprising: in response to determining that a location model is available for a present environment of a vehicle, generating a current model of the present environment using at least one sensor of the vehicle; isolating dynamic objects in the current model as a function of the location model, wherein the dynamic objects include semi-permanent objects of the present environment that change between the current model and the location model; and providing the dynamic objects to be identified and labeled, wherein the location model is a three-dimensional representation of the present environment that was previously acquired and labeled to identify the semi-permanent objects including at least static aspects of the present environment.
This invention relates to autonomous vehicle navigation, specifically identifying differences between a pre-existing location model and a current sensor-generated model of the vehicle's environment. The problem addressed is accurately distinguishing dynamic objects—including semi-permanent features—that change between the pre-existing model and real-time sensor data, which is critical for safe and reliable autonomous operation. The method begins by detecting the availability of a pre-existing location model for the vehicle's current environment. If available, the vehicle generates a current 3D model of the environment using onboard sensors. The system then isolates dynamic objects in the current model by comparing it to the pre-existing location model. Dynamic objects include semi-permanent features (e.g., construction barriers, temporary signs) that differ between the two models. These objects are then provided to a labeling system for identification and categorization. The pre-existing location model is a previously acquired and labeled 3D representation of the environment, including static aspects like roads, buildings, and permanent obstacles. The comparison process ensures the vehicle can adapt to environmental changes while maintaining accurate spatial awareness. This approach improves navigation safety by dynamically updating the vehicle's understanding of its surroundings.
15. The method of claim 14 , further comprising: in response to determining the location model is not available for the present environment, generating the location model by scanning the present environment using the at least one sensor to obtain three-dimensional data points representing present objects in the present environment that are collected to form the location model; and labeling the present objects to identify types of the present objects of the present environment.
This invention relates to generating and updating location models for environments using sensor data. The problem addressed is the lack of an available location model for a present environment, which is necessary for navigation, object recognition, or other spatial applications. The solution involves dynamically generating a location model when one is not available. The method uses at least one sensor to scan the present environment, collecting three-dimensional data points representing objects within the environment. These data points are processed to form a location model, which is a digital representation of the spatial layout and objects in the environment. Additionally, the objects in the environment are labeled to identify their types, enabling further analysis or interaction. The sensor data may include depth information, images, or other spatial measurements. The generated location model can then be used for tasks such as navigation, object tracking, or environmental mapping. This approach ensures that a location model is available even when a pre-existing one is not, improving the adaptability of systems operating in dynamic or unknown environments.
16. The method of claim 14 , wherein isolating the dynamic objects includes projecting the current model onto the location model to align the current model with the location model and to identify changes between the location model and the current model that correspond with the dynamic objects, and wherein isolating the dynamic objects includes labeling mobile objects that are part of the dynamic objects as provisional when the mobile objects are presently not moving but are determined to likely move and to label the dynamic objects that are semi-permanent objects according to an object class.
This invention relates to dynamic object detection and classification in environmental modeling, particularly for identifying and categorizing moving or potentially moving objects in a monitored space. The method involves comparing a current model of the environment with a pre-existing location model to detect changes, which are then identified as dynamic objects. The process includes projecting the current model onto the location model to align them and highlight differences, which correspond to dynamic objects. These objects are further classified into mobile objects and semi-permanent objects. Mobile objects that are stationary but likely to move are labeled as provisional, while semi-permanent objects are classified based on their object class. This approach improves environmental monitoring by distinguishing between transient and persistent dynamic elements, enhancing accuracy in tracking and predicting object behavior. The method is useful in applications such as surveillance, autonomous navigation, and environmental mapping, where distinguishing between static and dynamic elements is critical for decision-making.
17. The method of claim 16 , wherein projecting the current model onto the location model includes aligning a perspective of the current model with the location model to overlay corresponding points within the present environment, and wherein isolating the dynamic objects includes comparing the current model with the location model to produce observation data points that embody the dynamic objects of the present environment while removing static objects that form a background of the present environment that is unchanged.
This invention relates to environmental modeling and dynamic object detection, specifically for distinguishing moving objects from static backgrounds in a monitored environment. The method involves generating a location model representing the static background of the environment and a current model representing the environment at a later time. The current model is projected onto the location model by aligning their perspectives to overlay corresponding points, ensuring accurate spatial correlation. By comparing the current model with the location model, dynamic objects are isolated by identifying differences between the two models. This comparison produces observation data points that represent only the dynamic objects, effectively removing static background elements that remain unchanged. The technique enables real-time tracking of moving objects while suppressing irrelevant background information, improving accuracy in applications such as surveillance, autonomous navigation, and environmental monitoring. The method leverages spatial alignment and differential analysis to enhance object detection efficiency and reliability.
18. The method of claim 14 , further comprising: controlling the vehicle to navigate through the present environment according to the location model including updates derived from the dynamic objects, wherein the location model is a three-dimensional representation of the present environment that was previously acquired and labeled to identify static objects in the present environment.
This invention relates to autonomous vehicle navigation systems that improve real-time path planning by dynamically updating a pre-existing three-dimensional environmental model with detected dynamic objects. The system addresses the challenge of navigating through environments where static and dynamic objects coexist, ensuring safe and efficient movement by continuously refining the vehicle's understanding of its surroundings. The method involves using a pre-acquired and labeled three-dimensional representation of the environment, which identifies static objects such as buildings, roads, and obstacles. During operation, the vehicle detects and tracks dynamic objects like pedestrians, other vehicles, or moving obstacles. These dynamic objects are incorporated into the existing environmental model, allowing the vehicle to adjust its navigation path in real time. The system updates the model to reflect changes in the positions and movements of these dynamic objects, ensuring the vehicle can avoid collisions and optimize its route. By integrating dynamic object tracking with a pre-labeled static environment model, the invention enhances the vehicle's ability to navigate complex environments accurately and safely. This approach reduces reliance on real-time sensor data alone, improving robustness in varying conditions. The method ensures the vehicle maintains an up-to-date spatial awareness, enabling precise navigation even in dynamic and unpredictable scenarios.
19. The method of claim 14 , wherein providing the dynamic objects includes transmitting data from a LIDAR sensor and a camera sensor about each of the dynamic objects to a remote server for labeling, and wherein the data from the camera sensor is validation data for the dynamic objects.
This invention relates to systems for detecting and processing dynamic objects in an environment, such as for autonomous vehicle navigation or robotic operation. The problem addressed is the need for accurate and reliable identification of moving objects in real-time, which is critical for safe and efficient operation. The invention provides a method for detecting dynamic objects using sensor data, including LIDAR and camera inputs, and transmitting this data to a remote server for further processing and labeling. The LIDAR sensor captures spatial and positional data about the dynamic objects, while the camera sensor provides visual validation data to confirm the identity and characteristics of these objects. The remote server processes the combined sensor data to label the dynamic objects, ensuring accurate classification and tracking. This approach enhances the reliability of object detection by cross-referencing multiple sensor inputs and leveraging remote processing capabilities. The system is designed to improve the accuracy of dynamic object recognition in real-world applications, reducing errors and improving decision-making for autonomous systems. The use of both LIDAR and camera data ensures comprehensive coverage, with the camera providing visual context to validate the LIDAR-derived information. This method is particularly useful in environments where dynamic objects frequently appear, such as urban driving scenarios or industrial automation settings.
20. The method of claim 14 , wherein generating the current model includes scanning the present environment of the vehicle using at least a light detection and ranging (LIDAR) sensor to acquire the current model as a three-dimensional point cloud, and wherein providing the dynamic objects includes electronically communicating at least LiDAR data of the dynamic objects to a remote system for labeling to improve updating of the location model using the dynamic objects.
This invention relates to autonomous vehicle navigation systems that use environmental scanning and dynamic object tracking to improve location modeling. The system addresses the challenge of accurately mapping and updating a vehicle's surroundings in real-time, particularly when dealing with moving objects that can obstruct or alter the environment. The method involves generating a current model of the vehicle's environment by scanning it with a light detection and ranging (LiDAR) sensor, which captures the scene as a three-dimensional point cloud. This data is used to identify and track dynamic objects, such as other vehicles or pedestrians, within the environment. The system then communicates LiDAR data of these dynamic objects to a remote system for labeling, which helps refine and update the location model by incorporating information about the moving objects. This process enhances the accuracy and reliability of the vehicle's navigation and obstacle avoidance capabilities. The remote labeling system may use machine learning or other techniques to classify and track dynamic objects, ensuring the vehicle's model remains up-to-date with the latest environmental changes. The integration of remote data processing allows for continuous improvement of the vehicle's situational awareness, reducing the risk of navigation errors caused by unaccounted-for dynamic elements.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
April 24, 2017
November 26, 2019
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.